Today’s most advanced vision-language models (VLMs) remain proprietary. The strongest open-weight models rely heavily on synthetic data from proprietary VLMs to achieve good performance, effectively distilling these closed VLMs into open ones. As a result, the community has been missing foundational knowledge about how to build performant VLMs from scratch. We present Molmo, a new family of VLMs that are state-of-the-art in their class of openness. Our key contribution is a collection of new datasets called PixMo, including a dataset of highly detailed image captions for pre-training, a free-form image Q&A dataset for fine-tuning, and an innovative 2D pointing dataset, all collected without the use of external VLMs. The success of our approach relies on careful modeling choices, a well- tuned training pipeline, and, most critically, the quality of our newly collected datasets. Our best-in-class 72B model not only outperforms others in the class of open weight and data models, but also outperforms larger proprietary models including Claude 3.5 Sonnet, and Gemini 1.5 Pro and Flash, second only to GPT-4o based on both academic benchmarks and on a large human evaluation. Our model weights, new datasets, and source code are available at https://molmo.allenai.org/blog.

We present Segment Anything Model 2 (SAM 2), a foundation model towards solving promptable visual segmentation in images and videos. We build a data engine, which improves model and data via user interaction, to collect the largest video segmentation dataset to date. Our model is a simple transformer architecture with streaming memory for real-time video processing. SAM 2 trained on our data provides strong performance across a wide range of tasks. In video segmentation, we observe better accuracy, using 3x fewer interactions than prior approaches. In image segmentation, our model is more accurate and 6x faster than the Segment Anything Model (SAM). We believe that our data, model, and insights will serve as a significant milestone for video segmentation and related perception tasks. We are releasing our main model, the dataset, an interactive demo and code.

We introduce the Segment Anything (SA) project: a new task, model, and dataset for image segmentation. Using our efficient model in a data collection loop, we built the largest segmentation dataset to date (by far), with over 1 billion masks on 11M licensed and privacy respecting images. The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks. We evaluate its capabilities on numerous tasks and find that its zero-shot performance is impressive – often competitive with or even superior to prior fully supervised results. We are releasing the Segment Anything Model (SAM) and corresponding dataset (SA-1B) of 1B masks and 11M images at segment-anything.com to foster research into foundation models for computer vision. We recommend reading the full paper at: arxiv.org/abs/2304.02643.

This paper revisits the standard pretrain-then-finetune paradigm used in computer vision for visual recognition tasks. Typically, state-of-the-art foundation models are pretrained using large scale (weakly) supervised datasets with billions of images. We introduce an additional pre-pretraining stage that is simple and uses the self-supervised MAE technique to initialize the model. While MAE has only been shown to scale with the size of models, we find that it scales with the size of the training dataset as well. Thus, our MAE-based pre-pretraining scales with both model and data size making it applicable for training foundation models. Pre-pretraining consistently improves both the model convergence and the downstream transfer performance across a range of model scales (millions to billions of parameters), and dataset sizes (millions to billions of images). We measure the effectiveness of pre-pretraining on 10 different visual recognition tasks spanning image classification, video recognition, object detection, low-shot classification and zero-shot recognition. Our largest model achieves new state-of-the-art results on iNaturalist-18 (91.3%), 1-shot ImageNet-1k (62.1%), and zero-shot transfer on Food-101 (96.2%). Our study reveals that model initialization plays a significant role, even for web-scale pretraining with billions of images.

This paper shows that masked autoencoders (MAE) are scalable self-supervised learners for computer vision. Our MAE approach is simple: we mask random patches of the input image and reconstruct the missing pixels. It is based on two core designs. First, we develop an asymmetric encoder-decoder architecture, with an encoder that operates only on the visible subset of patches (without mask tokens), along with a lightweight decoder that reconstructs the original image from the latent representation and mask tokens. Second, we find that masking a high proportion of the input image, e.g., 75%, yields a nontrivial and meaningful self-supervisory task. Coupling these two designs enables us to train large models efficiently and effectively: we accelerate training (by 3× or more) and improve accuracy. Our scalable approach allows for learning high-capacity models that generalize well: e.g., a vanilla ViT-Huge model achieves the best accuracy (87.8%) among methods that use only ImageNet-1K data. Transfer performance in downstream tasks outperforms supervised pretraining and shows promising scaling behavior.

Model pre-training is a cornerstone of modern visual recognition systems. Although fully supervised pre-training on datasets like ImageNet is still the de-facto standard, recent studies suggest that large-scale weakly supervised pretraining can outperform fully supervised approaches. This paper revisits weakly-supervised pre-training of models using hashtag supervision with modern versions of residual networks and the largest-ever dataset of images and corresponding hashtags. We study the performance of the resulting models in various transfer-learning settings including zero-shot transfer. We also compare our models with those obtained via large-scale self-supervised learning. We find our weakly-supervised models to be very competitive across all settings, and find they substantially outperform their self-supervised counterparts. We also include an investigation into whether our models learned potentially troubling associations or stereotypes. Overall, our results provide a compelling argument for the use of weakly supervised learning in the development of visual recognition systems. Our models, Supervised Weakly through hashtAGs (SWAG), are available publicly.

We present Boundary IoU (Intersection-over-Union), a new segmentation evaluation measure focused on boundary quality. We perform an extensive analysis across different error types and object sizes and show that Boundary IoU is significantly more sensitive than the standard Mask IoU measure to boundary errors for large objects and does not over-penalize errors on smaller objects. The new quality measure displays several desirable characteristics like symmetry w.r.t. prediction/ground truth pairs and balanced responsiveness across scales, which makes it more suitable for segmentation evaluation than other boundary-focused measures like Trimap IoU and F-measure. Based on Boundary IoU, we update the standard evaluation protocols for instance and panoptic segmentation tasks by proposing the Boundary AP (Average Precision) and Boundary PQ (Panoptic Quality) metrics, respectively. Our experiments show that the new evaluation metrics track boundary quality improvements that are generally overlooked by current Mask IoU-based evaluation metrics. We hope that the adoption of the new boundary-sensitive evaluation metrics will lead to rapid progress in segmentation methods that improve boundary quality.1

In this work we analyze strategies for convolutional neural network scaling; that is, the process of scaling a base convolutional network to endow it with greater computational complexity and consequently representational power. Example scaling strategies may include increasing model width, depth, resolution, etc. While various scaling strategies exist, their tradeoffs are not fully understood. Existing analysis typically focuses on the interplay of accuracy and flops (floating point operations). Yet, as we demonstrate, various scaling strategies affect model parameters, activations, and consequently actual runtime quite differently. In our experiments we show the surprising result that numerous scaling strategies yield networks with similar accuracy but with widely varying properties. This leads us to propose a simple fast compound scaling strategy that encourages primarily scaling model width, while scaling depth and resolution to a lesser extent. Unlike currently popular scaling strategies, which result in about O(s) increase in model activation w.r.t. scaling flops by a factor of s, the proposed fast compound scaling results in close to $O\left( {\sqrt s } \right)$ increase in activations, while achieving excellent accuracy. Fewer activations leads to speedups on modern memory-bandwidth limited hardware (e.g., GPUs). More generally, we hope this work provides a framework for analyzing scaling strategies under various computational constraints.

We present a large-scale study on unsupervised spatiotemporal representation learning from videos. With a unified perspective on four recent image-based frameworks, we study a simple objective that can easily generalize all these methods to space-time. Our objective encourages temporally-persistent features in the same video, and in spite of its simplicity, it works surprisingly well across: (i) different unsupervised frameworks, (ii) pre-training datasets, (iii) downstream datasets, and (iv) backbone architectures. We draw a series of intriguing observations from this study, e.g., we discover that encouraging long-spanned persistency can be effective even if the timespan is 60 seconds. In addition to state-of-the-art results in multiple benchmarks, we report a few promising cases in which unsupervised pre-training can outperform its supervised counterpart. Code will be made available at https://github.com/facebookresearch/SlowFast.

We introduce PyTorchVideo, an open-source deep-learning library that provides a rich set of modular, efficient, and reproducible components for a variety of video understanding tasks, including classification, detection, self-supervised learning, and low-level processing. The library covers a full stack of video understanding tools including multimodal data loading, transformations, and models that reproduce state-of-the-art performance. PyTorchVideo further supports hardware acceleration that enables real-time inference on mobile devices. The library is based on PyTorch and can be used by any training framework; for example, PyTorchLightning, PySlowFast, or Classy Vision. PyTorchVideo is available at https://pytorchvideo.org/.

Vision transformer (ViT) models exhibit substandard optimizability. In particular, they are sensitive to the choice of optimizer (AdamW vs. SGD), optimizer hyperparameters, and training schedule length. In comparison, modern convolutional neural networks are easier to optimize. Why is this the case? In this work, we conjecture that the issue lies with the patchify stem of ViT models, which is implemented by a stride-p p×p convolution (p = 16 by default) applied to the input image. This large-kernel plus large-stride convolution runs counter to typical design choices of convolutional layers in neural networks. To test whether this atypical design choice causes an issue, we analyze the optimization behavior of ViT models with their original patchify stem versus a simple counterpart where we replace the ViT stem by a small number of stacked stride-two 3×3 convolutions. While the vast majority of computation in the two ViT designs is identical, we find that this small change in early visual processing results in markedly different training behavior in terms of the sensitivity to optimization settings as well as the final model accuracy. Using a convolutional stem in ViT dramatically increases optimization stability and also improves peak performance (by ∼1-2% top-1 accuracy on ImageNet-1k), while maintaining flops and runtime. The improvement can be observed across the wide spectrum of model complexities (from 1G to 36G flops) and dataset scales (from ImageNet-1k to ImageNet-21k). These findings lead us to recommend using a standard, lightweight convolutional stem for ViT models in this regime as a more robust architectural choice compared to the original ViT model design.

We present Momentum Contrast (MoCo) for unsupervised visual representation learning. From a perspective on contrastive learning as dictionary look-up, we build a dynamic dictionary with a queue and a moving-averaged encoder. This enables building a large and consistent dictionary on-the-fly that facilitates contrastive unsupervised learning. MoCo provides competitive results under the common linear protocol on ImageNet classification. More importantly, the representations learned by MoCo transfer well to downstream tasks. MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins. This suggests that the gap between unsupervised and supervised representation learning has been largely closed in many vision tasks.

We present a new method for efficient high-quality image segmentation of objects and scenes. By analogizing classical computer graphics methods for efficient rendering with over- and undersampling challenges faced in pixel labeling tasks, we develop a unique perspective of image segmentation as a rendering problem. From this vantage, we present the PointRend (Point-based Rendering) neural network module: a module that performs point-based segmentation predictions at adaptively selected locations based on an iterative subdivision algorithm. PointRend can be flexibly applied to both instance and semantic segmentation tasks by building on top of existing state-of-the-art models. While many concrete implementations of the general idea are possible, we show that a simple design already achieves excellent results. Qualitatively, PointRend outputs crisp object boundaries in regions that are over-smoothed by previous methods. Quantitatively, PointRend yields significant gains on COCO and Cityscapes, for both instance and semantic segmentation. PointRend's efficiency enables output resolutions that are otherwise impractical in terms of memory or computation compared to existing approaches. Code has been made available at https://github.com/facebookresearch/detectron2/tree/master/projects/PointRend.

In this work, we present a new network design paradigm. Our goal is to help advance the understanding of network design and discover design principles that generalize across settings. Instead of focusing on designing individual network instances, we design network design spaces that parametrize populations of networks. The overall process is analogous to classic manual design of networks, but elevated to the design space level. Using our methodology we explore the structure aspect of network design and arrive at a low-dimensional design space consisting of simple, regular networks that we call RegNet. The core insight of the RegNet parametrization is surprisingly simple: widths and depths of good networks can be explained by a quantized linear function. We analyze the RegNet design space and arrive at interesting findings that do not match the current practice of network design. The RegNet design space provides simple and fast networks that work well across a wide range of flop regimes. Under comparable training settings and flops, the RegNet models outperform the popular EfficientNet models while being up to 5x faster on GPUs.

Training competitive deep video models is an order of magnitude slower than training their counterpart image models. Slow training causes long research cycles, which hinders progress in video understanding research. Following standard practice for training image models, video model training has used a fixed mini-batch shape: a specific number of clips, frames, and spatial size. However, what is the optimal shape? High resolution models perform well, but train slowly. Low resolution models train faster, but are less accurate. Inspired by multigrid methods in numerical optimization, we propose to use variable mini-batch shapes with different spatial-temporal resolutions that are varied according to a schedule. The different shapes arise from resampling the training data on multiple sampling grids. Training is accelerated by scaling up the mini-batch size and learning rate when shrinking the other dimensions. We empirically demonstrate a general and robust grid schedule that yields a significant out-of-the-box training speedup without a loss in accuracy for different models (I3D, non-local, SlowFast), datasets (Kinetics, Something-Something, Charades), and training settings (with and without pre-training, 128 GPUs or 1 GPU). As an illustrative example, the proposed multigrid method trains a ResNet-50 SlowFast network 4.5x faster (wall-clock time, same hardware) while also improving accuracy (+0.8% absolute) on Kinetics-400 compared to baseline training. Code is available online.

Supervised ASR models have reached unprecedented levels of accuracy, thanks in part to ever-increasing amounts of labelled training data. However, in many applications and locales, only moderate amounts of data are available, which has led to a surge in semi- and weakly-supervised learning research. In this paper, we conduct a large-scale study evaluating the effectiveness of weakly-supervised learning for speech recognition by using loosely related contextual information as a surrogate for ground-truth labels. For weakly supervised training, we use 50k hours of public English social media videos along with their respective titles and post text to train an encoder-decoder transformer model. Our best encoder-decoder models achieve an average of 20.8% WER reduction over a 1000 hours supervised baseline, and an average of 13.4% WER reduction when using only the weakly supervised encoder for CTC fine-tuning. Our results show that our setup for weak supervision improved both the encoder acoustic representations as well as the decoder language generation abilities.

Many semi- and weakly-supervised approaches have been investigated for overcoming the labeling cost of building high quality speech recognition systems. On the challenging task of transcribing social media videos in low-resource conditions, we conduct a large scale systematic comparison between two self-labeling methods on one hand, and weakly-supervised pretraining using contextual metadata on the other. We investigate distillation methods at the frame level and the sequence level for hybrid, encoder-only CTC-based, and encoder-decoder speech recognition systems on Dutch and Romanian languages using 27,000 and 58,000 hours of unlabeled audio respectively. Although all approaches improved upon their respective baseline WERs by more than 8%, sequence-level distillation for encoder-decoder models provided the largest relative WER reduction of 20% compared to the strongest data-augmented supervised baseline.

We present a conceptually simple, flexible, and general framework for object instance segmentation. Our approach efficiently detects objects in an image while simultaneously generating a high-quality segmentation mask for each instance. The method, called Mask R-CNN, extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition. Mask R-CNN is simple to train and adds only a small overhead to Faster R-CNN, running at 5 fps. Moreover, Mask R-CNN is easy to generalize to other tasks, e.g., allowing us to estimate human poses in the same framework. We show top results in all three tracks of the COCO suite of challenges, including instance segmentation, bounding-box object detection, and person keypoint detection. Without bells and whistles, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners. We hope our simple and effective approach will serve as a solid baseline and help ease future research in instance-level recognition. Code has been made available at: https://github.com/facebookresearch/Detectron.

The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors. Code is at: https://github.com/facebookresearch/Detectron.

Progress on object detection is enabled by datasets that focus the research community’s attention on open challenges. This process led us from simple images to complex scenes and from bounding boxes to segmentation masks. In this work, we introduce LVIS (pronounced ‘el-vis’): a new dataset for Large Vocabulary Instance Segmentation. We plan to collect 2.2 million high-quality instance segmentation masks for over 1000 entry-level object categories in 164k images. Due to the Zipfian distribution of categories in natural images, LVIS naturally has a long tail of categories with few training samples. Given that state-of-the-art deep learning methods for object detection perform poorly in the low-sample regime, we believe that our dataset poses an important and exciting new scientific challenge. LVIS is available at http://www.lvisdataset.org.

The recently introduced panoptic segmentation task has renewed our community's interest in unifying the tasks of instance segmentation (for thing classes) and semantic segmentation (for stuff classes). However, current state-of-the-art methods for this joint task use separate and dissimilar networks for instance and semantic segmentation, without performing any shared computation. In this work, we aim to unify these methods at the architectural level, designing a single network for both tasks. Our approach is to endow Mask R-CNN, a popular instance segmentation method, with a semantic segmentation branch using a shared Feature Pyramid Network (FPN) backbone. Surprisingly, this simple baseline not only remains effective for instance segmentation, but also yields a lightweight, top-performing method for semantic segmentation. In this work, we perform a detailed study of this minimally extended version of Mask R-CNN with FPN, which we refer to as Panoptic FPN, and show it is a robust and accurate baseline for both tasks. Given its effectiveness and conceptual simplicity, we hope our method can serve as a strong baseline and aid future research in panoptic segmentation.

We propose and study a task we name panoptic segmentation (PS). Panoptic segmentation unifies the typically distinct tasks of semantic segmentation (assign a class label to each pixel) and instance segmentation (detect and segment each object instance). The proposed task requires generating a coherent scene segmentation that is rich and complete, an important step toward real-world vision systems. While early work in computer vision addressed related image/scene parsing tasks, these are not currently popular, possibly due to lack of appropriate metrics or associated recognition challenges. To address this, we propose a novel panoptic quality (PQ) metric that captures performance for all classes (stuff and things) in an interpretable and unified manner. Using the proposed metric, we perform a rigorous study of both human and machine performance for PS on three existing datasets, revealing interesting insights about the task. The aim of our work is to revive the interest of the community in a more unified view of image segmentation. For more analysis and up-to-date results, please check the arXiv version of the paper: https://arxiv.org/abs/1801.00868.

To understand the world, we humans constantly need to relate the present to the past, and put events in context. In this paper, we enable existing video models to do the same. We propose a long-term feature bank—supportive information extracted over the entire span of a video—to augment state-of-the-art video models that otherwise would only view short clips of 2-5 seconds. Our experiments demonstrate that augmenting 3D convolutional networks with a long-term feature bank yields state-of-the-art results on three challenging video datasets: AVA, EPIC-Kitchens, and Charades. Code is available online.

Sliding-window object detectors that generate bounding-box object predictions over a dense, regular grid have advanced rapidly and proven popular. In contrast, modern instance segmentation approaches are dominated by methods that first detect object bounding boxes, and then crop and segment these regions, as popularized by Mask R-CNN. In this work, we investigate the paradigm of dense sliding-window instance segmentation, which is surprisingly under-explored. Our core observation is that this task is fundamentally different than other dense prediction tasks such as semantic segmentation or bounding-box object detection, as the output at every spatial location is itself a geometric structure with its own spatial dimensions. To formalize this, we treat dense instance segmentation as a prediction task over 4D tensors and present a general framework called TensorMask that explicitly captures this geometry and enables novel operators on 4D tensors. We demonstrate that the tensor view leads to large gains over baselines that ignore this structure, and leads to results comparable to Mask R-CNN. These promising results suggest that TensorMask can serve as a foundation for novel advances in dense mask prediction and a more complete understanding of the task. Code will be made available.

We report competitive results on object detection and instance segmentation on the COCO dataset using standard models trained from random initialization. The results are no worse than their ImageNet pre-training counterparts even when using the hyper-parameters of the baseline system (Mask R-CNN) that were optimized for fine-tuning pre-trained models, with the sole exception of increasing the number of training iterations so the randomly initialized models may converge. Training from random initialization is surprisingly robust; our results hold even when: (i) using only 10% of the training data, (ii) for deeper and wider models, and (iii) for multiple tasks and metrics. Experiments show that ImageNet pre-training speeds up convergence early in training, but does not necessarily provide regularization or improve final target task accuracy. To push the envelope we demonstrate 50.9 AP on COCO object detection without using any external data-a result on par with the top COCO 2017 competition results that used ImageNet pre-training. These observations challenge the conventional wisdom of ImageNet pre-training for dependent tasks and we expect these discoveries will encourage people to rethink the current de facto paradigm of `pretraining and fine-tuning' in computer vision.

Neural networks for image recognition have evolved through extensive manual design from simple chain-like models to structures with multiple wiring paths. The success of ResNets and DenseNets is due in large part to their innovative wiring plans. Now, neural architecture search (NAS) studies are exploring the joint optimization of wiring and operation types, however, the space of possible wirings is constrained and still driven by manual design despite being searched. In this paper, we explore a more diverse set of connectivity patterns through the lens of randomly wired neural networks. To do this, we first define the concept of a stochastic network generator that encapsulates the entire network generation process. Encapsulation provides a unified view of NAS and randomly wired networks. Then, we use three classical random graph models to generate randomly wired graphs for networks. The results are surprising: several variants of these random generators yield network instances that have competitive accuracy on the ImageNet benchmark. These results suggest that new efforts focusing on designing better network generators may lead to new breakthroughs by exploring less constrained search spaces with more room for novel design.

Understanding and reasoning about physics is an important ability of intelligent agents. We develop the PHYRE benchmark for physical reasoning that contains a set of simple classical mechanics puzzles in a 2D physical environment. The benchmark is designed to encourage the development of learning algorithms that are sample-efficient and generalize well across puzzles. We test several modern learning algorithms on PHYRE and find that these algorithms fall short in solving the puzzles efficiently. We expect that PHYRE will encourage the development of novel sample-efficient agents that learn efficient but useful models of physics. For code and to play PHYRE for yourself, please visit https://player.phyre.ai.

Both convolutional and recurrent operations are building blocks that process one local neighborhood at a time. In this paper, we present non-local operations as a generic family of building blocks for capturing long-range dependencies. Inspired by the classical non-local means method in computer vision, our non-local operation computes the response at a position as a weighted sum of the features at all positions. This building block can be plugged into many computer vision architectures. On the task of video classification, even without any bells and whistles, our non-local models can compete or outperform current competition winners on both Kinetics and Charades datasets. In static image recognition, our non-local models improve object detection/segmentation and pose estimation on the COCO suite of tasks. Code is available at https://github.com/facebookresearch/video-nonlocal-net .

To understand the visual world, a machine must not only recognize individual object instances but also how they interact. Humans are often at the center of such interactions and detecting human-object interactions is an important practical and scientific problem. In this paper, we address the task of detecting (human, verb, object) triplets in challenging everyday photos. We propose a novel model that is driven by a human-centric approach. Our hypothesis is that the appearance of a person - their pose, clothing, action - is a powerful cue for localizing the objects they are interacting with. To exploit this cue, our model learns to predict an action-specific density over target object locations based on the appearance of a detected person. Our model also jointly learns to detect people and objects, and by fusing these predictions it efficiently infers interaction triplets in a clean, jointly trained end-to-end system we call InteractNet. We validate our approach on the recently introduced Verbs in COCO (V-COCO) and HICO-DET datasets, where we show quantitatively compelling results.

Most methods for object instance segmentation require all training examples to be labeled with segmentation masks. This requirement makes it expensive to annotate new categories and has restricted instance segmentation models to ~100 well-annotated classes. The goal of this paper is to propose a new partially supervised training paradigm, together with a novel weight transfer function, that enables training instance segmentation models on a large set of categories all of which have box annotations, but only a small fraction of which have mask annotations. These contributions allow us to train Mask R-CNN to detect and segment 3000 visual concepts using box annotations from the Visual Genome dataset and mask annotations from the 80 classes in the COCO dataset. We evaluate our approach in a controlled study on the COCO dataset. This work is a first step towards instance segmentation models that have broad comprehension of the visual world.

We introduce an interactive learning framework for the development and testing of intelligent visual systems, called learning-by-asking (LBA). We explore LBA in context of the Visual Question Answering (VQA) task. LBA differs from standard VQA training in that most questions are not observed during training time, and the learner must ask questions it wants answers to. Thus, LBA more closely mimics natural learning and has the potential to be more data-efficient than the traditional VQA setting. We present a model that performs LBA on the CLEVR dataset, and show that it automatically discovers an easy-to-hard curriculum when learning interactively from an oracle. Our LBA generated data consistently matches or outperforms the CLEVR train data and is more sample efficient. We also show that our model asks questions that generalize to state-of-the-art VQA models and to novel test time distributions.

We investigate omni-supervised learning, a special regime of semi-supervised learning in which the learner exploits all available labeled data plus internet-scale sources of unlabeled data. Omni-supervised learning is lower-bounded by performance on existing labeled datasets, offering the potential to surpass state-of-the-art fully supervised methods. To exploit the omni-supervised setting, we propose data distillation, a method that ensembles predictions from multiple transformations of unlabeled data, using a single model, to automatically generate new training annotations. We argue that visual recognition models have recently become accurate enough that it is now possible to apply classic ideas about self-training to challenging real-world data. Our experimental results show that in the cases of human keypoint detection and general object detection, state-of-the-art models trained with data distillation surpass the performance of using labeled data from the COCO dataset alone.

Humans can quickly learn new visual concepts, perhaps because they can easily visualize or imagine what novel objects look like from different views. Incorporating this ability to hallucinate novel instances of new concepts might help machine vision systems perform better low-shot learning, i.e., learning concepts from few examples. We present a novel approach to low-shot learning that uses this idea. Our approach builds on recent progress in meta-learning ("learning to learn") by combining a meta-learner with a "hallucinator" that produces additional training examples, and optimizing both models jointly. Our hallucinator can be incorporated into a variety of meta-learners and provides significant gains: up to a 6 point boost in classification accuracy when only a single training example is available, yielding state-of-the-art performance on the challenging ImageNet low-shot classification benchmark.

When building artificial intelligence systems that can reason and answer questions about visual data, we need diagnostic tests to analyze our progress and discover short-comings. Existing benchmarks for visual question answering can help, but have strong biases that models can exploit to correctly answer questions without reasoning. They also conflate multiple sources of error, making it hard to pinpoint model weaknesses. We present a diagnostic dataset that tests a range of visual reasoning abilities. It contains minimal biases and has detailed annotations describing the kind of reasoning each question requires. We use this dataset to analyze a variety of modern visual reasoning systems, providing novel insights into their abilities and limitations.

Feature pyramids are a basic component in recognition systems for detecting objects at different scales. But pyramid representations have been avoided in recent object detectors that are based on deep convolutional networks, partially because they are slow to compute and memory intensive. In this paper, we exploit the inherent multi-scale, pyramidal hierarchy of deep convolutional networks to construct feature pyramids with marginal extra cost. A top-down architecture with lateral connections is developed for building high-level semantic feature maps at all scales. This architecture, called a Feature Pyramid Network (FPN), shows significant improvement as a generic feature extractor in several applications. Using a basic Faster R-CNN system, our method achieves state-of-the-art single-model results on the COCO detection benchmark without bells and whistles, surpassing all existing single-model entries including those from the COCO 2016 challenge winners. In addition, our method can run at 5 FPS on a GPU and thus is a practical and accurate solution to multi-scale object detection. Code will be made publicly available.

This paper presents a novel yet intuitive approach to unsupervised feature learning. Inspired by the human visual system, we explore whether low-level motion-based grouping cues can be used to learn an effective visual representation. Specifically, we use unsupervised motion-based segmentation on videos to obtain segments, which we use as pseudo ground truth to train a convolutional network to segment objects from a single frame. Given the extensive evidence that motion plays a key role in the development of the human visual system, we hope that this straightforward approach to unsupervised learning will be more effective than cleverly designed pretext tasks studied in the literature. Indeed, our extensive experiments show that this is the case. When used for transfer learning on object detection, our representation significantly outperforms previous unsupervised approaches across multiple settings, especially when training data for the target task is scarce.

We present a simple, highly modularized network architecture for image classification. Our network is constructed by repeating a building block that aggregates a set of transformations with the same topology. Our simple design results in a homogeneous, multi-branch architecture that has only a few hyper-parameters to set. This strategy exposes a new dimension, which we call cardinality (the size of the set of transformations), as an essential factor in addition to the dimensions of depth and width. On the ImageNet-1K dataset, we empirically show that even under the restricted condition of maintaining complexity, increasing cardinality is able to improve classification accuracy. Moreover, increasing cardinality is more effective than going deeper or wider when we increase the capacity. Our models, named ResNeXt, are the foundations of our entry to the ILSVRC 2016 classification task in which we secured 2nd place. We further investigate ResNeXt on an ImageNet-5K set and the COCO detection set, also showing better results than its ResNet counterpart. The code and models are publicly available online.

Low-shot visual learning-the ability to recognize novel object categories from very few examples-is a hallmark of human visual intelligence. Existing machine learning approaches fail to generalize in the same way. To make progress on this foundational problem, we present a low-shot learning benchmark on complex images that mimics challenges faced by recognition systems in the wild. We then propose (1) representation regularization techniques, and (2) techniques to hallucinate additional training examples for data-starved classes. Together, our methods improve the effectiveness of convolutional networks in low-shot learning, improving the one-shot accuracy on novel classes by 2.3× on the challenging ImageNet dataset.

We present a conceptually simple, flexible, and general framework for object instance segmentation. Our approach efficiently detects objects in an image while simultaneously generating a high-quality segmentation mask for each instance. The method, called Mask R-CNN, extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition. Mask R-CNN is simple to train and adds only a small overhead to Faster R-CNN, running at 5 fps. Moreover, Mask R-CNN is easy to generalize to other tasks, e.g., allowing us to estimate human poses in the same framework. We show top results in all three tracks of the COCO suite of challenges, including instance segmentation, bounding-box object detection, and person keypoint detection. Without tricks, Mask R-CNN outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners. We hope our simple and effective approach will serve as a solid baseline and help ease future research in instance-level recognition. Code will be made available.

Existing methods for visual reasoning attempt to directly map inputs to outputs using black-box architectures without explicitly modeling the underlying reasoning processes. As a result, these black-box models often learn to exploit biases in the data rather than learning to perform visual reasoning. Inspired by module networks, this paper proposes a model for visual reasoning that consists of a program generator that constructs an explicit representation of the reasoning process to be performed, and an execution engine that executes the resulting program to produce an answer. Both the program generator and the execution engine are implemented by neural networks, and are trained using a combination of backpropagation and REINFORCE. Using the CLEVR benchmark for visual reasoning, we show that our model significantly outperforms strong baselines and generalizes better in a variety of settings.

The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors.

Recognition algorithms based on convolutional networks (CNNs) typically use the output of the last layer as a feature representation. However, the information in this layer may be too coarse spatially to allow precise localization. On the contrary, earlier layers may be precise in localization but will not capture semantics. To get the best of both worlds, we define the hypercolumn at a pixel as the vector of activations of all CNN units above that pixel. Using hypercolumns as pixel descriptors, we show results on three fine-grained localization tasks: simultaneous detection and segmentation, where we improve state-of-the-art from 49.7 mean APr to 62.4, keypoint localization, where we get a 3.3 point boost over a strong regression baseline using CNN features, and part labeling, where we show a 6.6 point gain over a strong baseline.

State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet [1] and Fast R-CNN [2] have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features-using the recently popular terminology of neural networks with 'attention' mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model [3] , our detection system has a frame rate of 5 fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available.

Most object detectors contain two important components: a feature extractor and an object classifier. The feature extractor has rapidly evolved with significant research efforts leading to better deep convolutional architectures. The object classifier, however, has not received much attention and many recent systems (like SPPnet and Fast/Faster R-CNN) use simple multi-layer perceptrons. This paper demonstrates that carefully designing deep networks for object classification is just as important. We experiment with region-wise classifier networks that use shared, region-independent convolutional features. We call them "Networks on Convolutional feature maps" (NoCs). We discover that aside from deep feature maps, a deep and convolutional per-region classifier is of particular importance for object detection, whereas latest superior image classification models (such as ResNets and GoogLeNets) do not directly lead to good detection accuracy without using such a per-region classifier. We show by experiments that despite the effective ResNets and Faster R-CNN systems, the design of NoCs is an essential element for the 1st-place winning entries in ImageNet and MS COCO challenges 2015.

It is well known that contextual and multi-scale representations are important for accurate visual recognition. In this paper we present the Inside-Outside Net (ION), an object detector that exploits information both inside and outside the region of interest. Contextual information outside the region of interest is integrated using spatial recurrent neural networks. Inside, we use skip pooling to extract information at multiple scales and levels of abstraction. Through extensive experiments we evaluate the design space and provide readers with an overview of what tricks of the trade are important. ION improves state-of-the-art on PASCAL VOC 2012 object detection from 73.9% to 77.9% mAP. On the new and more challenging MS COCO dataset, we improve state-of-the-art from 19.7% to 33.1% mAP. In the 2015 MS COCO Detection Challenge, our ION model won "Best Student Entry" and finished 3rd place overall. As intuition suggests, our detection results provide strong evidence that context and multi-scale representations improve small object detection.

When human annotators are given a choice about what to label in an image, they apply their own subjective judgments on what to ignore and what to mention. We refer to these noisy "human-centric" annotations as exhibiting human reporting bias. Examples of such annotations include image tags and keywords found on photo sharing sites, or in datasets containing image captions. In this paper, we use these noisy annotations for learning visually correct image classifiers. Such annotations do not use consistent vocabulary, and miss a significant amount of the information present in an image, however, we demonstrate that the noise in these annotations exhibits structure and can be modeled. We propose an algorithm to decouple the human reporting bias from the correct visually grounded labels. Our results are highly interpretable for reporting "what's in the image" versus "what's worth saying." We demonstrate the algorithm's efficacy along a variety of metrics and datasets, including MS COCO and Yahoo Flickr 100M.We show significant improvements over traditional algorithms for both image classification and image captioning, doubling the performance of existing methods in some cases.

We present YOLO, a new approach to object detection. Prior work on object detection repurposes classifiers to perform detection. Instead, we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast. Our base YOLO model processes images in real-time at 45 frames per second. A smaller version of the network, Fast YOLO, processes an astounding 155 frames per second while still achieving double the mAP of other real-time detectors. Compared to state-of-the-art detection systems, YOLO makes more localization errors but is less likely to predict false positives on background. Finally, YOLO learns very general representations of objects. It outperforms other detection methods, including DPM and R-CNN, when generalizing from natural images to other domains like artwork.

The field of object detection has made significant advances riding on the wave of region-based ConvNets, but their training procedure still includes many heuristics and hyperparameters that are costly to tune. We present a simple yet surprisingly effective online hard example mining (OHEM) algorithm for training region-based ConvNet detectors. Our motivation is the same as it has always been - detection datasets contain an overwhelming number of easy examples and a small number of hard examples. Automatic selection of these hard examples can make training more effective and efficient. OHEM is a simple and intuitive algorithm that eliminates several heuristics and hyperparameters in common use. But more importantly, it yields consistent and significant boosts in detection performance on benchmarks like PASCAL VOC 2007 and 2012. Its effectiveness increases as datasets become larger and more difficult, as demonstrated by the results on the MS COCO dataset. Moreover, combined with complementary advances in the field, OHEM leads to state-of-the-art results of 78.9% and 76.3% mAP on PASCAL VOC 2007 and 2012 respectively.

Clustering is central to many data-driven application domains and has been studied extensively in terms of distance functions and grouping algorithms. Relatively little work has focused on learning representations for clustering. In this paper, we propose Deep Embedded Clustering (DEC), a method that simultaneously learns feature representations and cluster assignments using deep neural networks. DEC learns a mapping from the data space to a lower-dimensional feature space in which it iteratively optimizes a clustering objective. Our experimental evaluations on image and text corpora show significant improvement over state-of-the-art methods.

Ting-Hao Kenneth Huang, Francis Ferraro, Nasrin Mostafazadeh, Ishan Misra, Aishwarya Agrawal, Jacob Devlin, Ross Girshick, Xiaodong He, Pushmeet Kohli, Dhruv Batra, C. Lawrence Zitnick, Devi Parikh, Lucy Vanderwende, Michel Galley, Margaret Mitchell. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.

Object detection performance, as measured on the canonical PASCAL VOC Challenge datasets, plateaued in the final years of the competition. The best-performing methods were complex ensemble systems that typically combined multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 50 percent relative to the previous best result on VOC 2012-achieving a mAP of 62.4 percent. Our approach combines two ideas: (1) one can apply high-capacity convolutional networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data are scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, boosts performance significantly. Since we combine region proposals with CNNs, we call the resulting model an R-CNN or Region-based Convolutional Network. Source code for the complete system is available at http://www.cs.berkeley.edu/~rbg/rcnn.

Deformable part models (DPMs) and convolutional neural networks (CNNs) are two widely used tools for visual recognition. They are typically viewed as distinct approaches: DPMs are graphical models (Markov random fields), while CNNs are “black-box” non-linear classifiers. In this paper, we show that a DPM can be formulated as a CNN, thus providing a synthesis of the two ideas. Our construction involves unrolling the DPM inference algorithm and mapping each step to an equivalent CNN layer. From this perspective, it is natural to replace the standard image features used in DPMs with a learned feature extractor. We call the resulting model a DeepPyramid DPM and experimentally validate it on PASCAL VOC object detection. We find that DeepPyramid DPMs significantly outperform DPMs based on histograms of oriented gradients features (HOG) and slightly outperforms a comparable version of the recently introduced R-CNN detection system, while running significantly faster.

The goal of this work is to represent objects in an RGB-D scene with corresponding 3D models from a library. We approach this problem by first detecting and segmenting object instances in the scene and then using a convolutional neural network (CNN) to predict the pose of the object. This CNN is trained using pixel surface normals in images containing renderings of synthetic objects. When tested on real data, our method outperforms alternative algorithms trained on real data. We then use this coarse pose estimate along with the inferred pixel support to align a small number of prototypical models to the data, and place into the scene the model that fits best. We observe a 48% relative improvement in performance at the task of 3D detection over the current state-of-the-art [34], while being an order of magnitude faster.

Recognition algorithms based on convolutional networks (CNNs) typically use the output of the last layer as a feature representation. However, the information in this layer may be too coarse spatially to allow precise localization. On the contrary, earlier layers may be precise in localization but will not capture semantics. To get the best of both worlds, we define the hypercolumn at a pixel as the vector of activations of all CNN units above that pixel. Using hypercolumns as pixel descriptors, we show results on three fine-grained localization tasks: simultaneous detection and segmentation [22], where we improve state-of-the-art from 49.7 mean APr[22] to 60.0, keypoint localization, where we get a 3.3 point boost over [20], and part labeling, where we show a 6.6 point gain over a strong baseline.

This paper proposes a Fast Region-based Convolutional Network method (Fast R-CNN) for object detection. Fast R-CNN builds on previous work to efficiently classify object proposals using deep convolutional networks. Compared to previous work, Fast R-CNN employs several innovations to improve training and testing speed while also increasing detection accuracy. Fast R-CNN trains the very deep VGG16 network 9x faster than R-CNN, is 213x faster at test-time, and achieves a higher mAP on PASCAL VOC 2012. Compared to SPPnet, Fast R-CNN trains VGG16 3x faster, tests 10x faster, and is more accurate. Fast R-CNN is implemented in Python and C++ (using Caffe) and is available under the open-source MIT License at https://github.com/rbgirshick/fast-rcnn.

There are multiple cues in an image which reveal what action a person is performing. For example, a jogger has a pose that is characteristic for jogging, but the scene (e.g. road, trail) and the presence of other joggers can be an additional source of information. In this work, we exploit the simple observation that actions are accompanied by contextual cues to build a strong action recognition system. We adapt RCNN to use more than one region for classification while still maintaining the ability to localize the action. We call our system R*CNN. The action-specific models and the feature maps are trained jointly, allowing for action specific representations to emerge. R*CNN achieves 90.2% mean AP on the PASAL VOC Action dataset, outperforming all other approaches in the field by a significant margin. Last, we show that R*CNN is not limited to action recognition. In particular, R*CNN can also be used to tackle fine-grained tasks such as attribute classification. We validate this claim by reporting state-of-the-art performance on the Berkeley Attributes of People dataset.

We investigate the importance of parts for the tasks of action and attribute classification. We develop a part-based approach by leveraging convolutional network features inspired by recent advances in computer vision. Our part detectors are a deep version of poselets and capture parts of the human body under a distinct set of poses. For the tasks of action and attribute classification, we train holistic convolutional neural networks and show that adding parts leads to top-performing results for both tasks. We observe that for deeper networks parts are less significant. In addition, we demonstrate the effectiveness of our approach when we replace an oracle person detector, as is the default in the current evaluation protocol for both tasks, with a state-of-the-art person detection system.

State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully-convolutional network that simultaneously predicts object bounds and objectness scores at each position. RPNs are trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. With a simple alternating optimization, RPN and Fast R-CNN can be trained to share convolutional features. For the very deep VGG-16 model, our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007 (73.2% mAP) and 2012 (70.4% mAP) using 300 proposals per image. Code is available at https://github.com/ShaoqingRen/faster_rcnn.

The problem of real-time multiclass object recognition is of great practical importance in object recognition. In this paper, we describe a framework that simultaneously utilizes shared representation, reconstruction sparsity, and parallelism to enable real-time multiclass object detection with deformable part models at 5Hz on a laptop computer with almost no decrease in task performance. Our framework is trained in the standard structured output prediction formulation and is generically applicable for speeding up object recognition systems where the computational bottleneck is in multiclass, multi-convolutional inference. We experimentally demonstrate the efficiency and task performance of our method on PASCAL VOC, subset of ImageNet, Caltech101 and Caltech256 dataset.

Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30% relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3%. Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http://www.cs.berkeley.edu/~rbg/rcnn.

A k-poselet is a deformable part model (DPM) with k parts, where each of the parts is a poselet, aligned to a specific configuration of keypoints based on ground-truth annotations. A separate template is used to learn the appearance of each part. The parts are allowed to move with respect to each other with a deformation cost that is learned at training time. This model is richer than both the traditional version of poselets and DPMs. It enables a unified approach to person detection and keypoint prediction which, barring contemporaneous approaches based on CNN features, achieves state-of-the-art keypoint prediction while maintaining competitive detection performance.

We study the problem of understanding objects in detail, intended as recognizing a wide array of fine-grained object attributes. To this end, we introduce a dataset of 7, 413 airplanes annotated in detail with parts and their attributes, leveraging images donated by airplane spotters and crowd-sourcing both the design and collection of the detailed annotations. We provide a number of insights that should help researchers interested in designing fine-grained datasets for other basic level categories. We show that the collected data can be used to study the relation between part detection and attribute prediction by diagnosing the performance of classifiers that pool information from different parts of an object. We note that the prediction of certain attributes can benefit substantially from accurate part detection. We also show that, differently from previous results in object detection, employing a large number of part templates can improve detection accuracy at the expenses of detection speed. We finally propose a coarse-to-fine approach to speed up detection through a hierarchical cascade algorithm.

Learning to localize objects with minimal supervision is an important problem in computer vision, since large fully annotated datasets are extremely costly to obtain. In this paper, we propose a new method that achieves this goal with only image-level labels of whether the objects are present or not. Our approach combines a discriminative submodular cover problem for automatically discovering a set of positive object windows with a smoothed latent SVM formulation. The latter allows us to leverage efficient quasi-Newton optimization techniques. Our experiments demonstrate that the proposed approach provides a 50% relative improvement in mean average precision over the current state-of-the-art on PASCAL VOC 2007 detection.

Caffe provides multimedia scientists and practitioners with a clean and modifiable framework for state-of-the-art deep learning algorithms and a collection of reference models. The framework is a BSD-licensed C++ library with Python and MATLAB bindings for training and deploying general-purpose convolutional neural networks and other deep models efficiently on commodity architectures. Caffe fits industry and internet-scale media needs by CUDA GPU computation, processing over 40 million images a day on a single K40 or Titan GPU (approx 2 ms per image). By separating model representation from actual implementation, Caffe allows experimentation and seamless switching among platforms for ease of development and deployment from prototyping machines to cloud environments.

In this paper, we show how to train a deformable part model (DPM) fast-typically in less than 20 minutes, or four times faster than the current fastest method-while maintaining high average precision on the PASCAL VOC datasets. At the core of our approach is "latent LDA," a novel generalization of linear discriminant analysis for learning latent variable models. Unlike latent SVM, latent LDA uses efficient closed-form updates and does not require an expensive search for hard negative examples. Our approach also acts as a springboard for a detailed experimental study of DPM training. We isolate and quantify the impact of key training factors for the first time (e.g., How important are discriminative SVM filters? How important is joint parameter estimation? How many negative images are needed for training?). Our findings yield useful insights for researchers working with Markov random fields and part-based models, and have practical implications for speeding up tasks such as model selection.

Shared representations are highly appealing due to their potential for gains in computational and statistical efficiency. Compressing a shared representation leads to greater computational savings, but at the same time can severely decrease performance on a target task. Recently, sparselets (Song et al., 2012) were introduced as a new shared intermediate representation for multiclass object detection with deformable part models (Felzenszwalb et al., 2010a), showing significant speedup factors, but with a large decrease in task performance. In this paper we describe a new training framework that learns which sparselets to activate in order to optimize a discriminative objective, leading to larger speedup factors with no decrease in task performance. We first reformulate sparselets in a general structured output prediction framework, then analyze when sparselets lead to computational efficiency gains, and lastly show experimental results on object detection and image classification tasks. Our experimental results demonstrate that discriminative activation substantially outperforms the previous reconstructive approach which, together with our structured output prediction formulation, make sparselets broadly applicable and significantly more effective.

We describe two new approaches to human pose estimation. Both can quickly and accurately predict the 3D positions of body joints from a single depth image without using any temporal information. The key to both approaches is the use of a large, realistic, and highly varied synthetic set of training images. This allows us to learn models that are largely invariant to factors such as pose, body shape, field-of-view cropping, and clothing. Our first approach employs an intermediate body parts representation, designed so that an accurate per-pixel classification of the parts will localize the joints of the body. The second approach instead directly regresses the positions of body joints. By using simple depth pixel comparison features and parallelizable decision forests, both approaches can run super-real time on consumer hardware. Our evaluation investigates many aspects of our methods, and compares the approaches to each other and to the state of the art. Results on silhouettes suggest broader applicability to other imaging modalities.

We present a new approach to general-activity human pose estimation from depth images, building on Hough forests. We extend existing techniques in several ways: real time prediction of multiple 3D joints, explicit learning of voting weights, vote compression to allow larger training sets, and a comparison of several decision-tree training objectives. Key aspects of our work include: regression directly from the raw depth image, without the use of an arbitrary intermediate representation; applicability to general motions (not constrained to particular activities) and the ability to localize occluded as well as visible body joints. Experimental results demonstrate that our method produces state of the art results on several data sets including the challenging MSRC-5000 pose estimation test set, at a speed of about 200 frames per second. Results on silhouettes suggest broader applicability to other imaging modalities.

Compositional models provide an elegant formalism for representing the visual appearance of highly variable objects. While such models are appealing from a theoretical point of view, it has been difficult to demonstrate that they lead to performance advantages on challenging datasets. Here we develop a grammar model for person detection and show that it outperforms previous high-performance systems on the PASCAL benchmark. Our model represents people using a hierarchy of deformable parts, variable structure and an explicit model of occlusion for partially visible objects. To train the model, we introduce a new discriminative framework for learning structured prediction models from weakly-labeled data.

We describe a general method for building cascade classifiers from part-based deformable models such as pictorial structures. We focus primarily on the case of star-structured models and show how a simple algorithm based on partial hypothesis pruning can speed up object detection by more than one order of magnitude without sacrificing detection accuracy. In our algorithm, partial hypotheses are pruned with a sequence of thresholds. In analogy to probably approximately correct (PAC) learning, we introduce the notion of probably approximately admissible (PAA) thresholds. Such thresholds provide theoretical guarantees on the performance of the cascade method and can be computed from a small sample of positive examples. Finally, we outline a cascade detection algorithm for a general class of models defined by a grammar formalism. This class includes not only tree-structured pictorial structures but also richer models that can represent each part recursively as a mixture of other parts.

We describe an object detection system based on mixtures of multiscale deformable part models. Our system is able to represent highly variable object classes and achieves state-of-the-art results in the PASCAL object detection challenges. While deformable part models have become quite popular, their value had not been demonstrated on difficult benchmarks such as the PASCAL data sets. Our system relies on new methods for discriminative training with partially labeled data. We combine a margin-sensitive approach for data-mining hard negative examples with a formalism we call latent SVM. A latent SVM is a reformulation of MI--SVM in terms of latent variables. A latent SVM is semiconvex, and the training problem becomes convex once latent information is specified for the positive examples. This leads to an iterative training algorithm that alternates between fixing latent values for positive examples and optimizing the latent SVM objective function.

To recognize three-dimensional objects it is important to model how their appearances can change due to changes in viewpoint. A key aspect of this involves understanding which object features can be simultaneously visible under different viewpoints. We address this problem in an image-based framework, in which we use a limited number of images of an object taken from unknown viewpoints to determine which subsets of features might be simultaneously visible in other views. This leads to the problem of determining whether a set of images, each containing a set of features, is consistent with a single 3D object. We assume that each feature is visible from a disk of viewpoints on the viewing sphere. In this case we show the problem is NP-hard in general, but can be solved efficiently when all views come from a circle on the viewing sphere. We also give iterative algorithms that can handle noisy data and converge to locally optimal solutions in the general case. Our techniques can also be used to recover viewpoint information from the set of features that are visible in different images. We show that these algorithms perform well both on synthetic data and images from the COIL dataset.